Forecasting high-frequency currency tail risks with EVT and technical indicators across pandemic phases

This research employs Extreme Value Theory (EVT) to forecast high-frequency tail risks in USD/INR currency returns during the COVID-19 period. Unlike traditional studies that focus on lower-frequency data, we leverage high-frequency returns to capture rapid market adjustments and tail risk dynamics with greater precision. We decompose hourly returns into extreme positive (EPR) and negative (ENR) returns, revealing structural shifts in volatility, market efficiency, and lead-lag dynamics across pandemic phases. Our findings indicate that the pandemic disrupted random walk, as prior-hour losses (gains) significantly predicted subsequent gains (losses), highlighting trend persistence and overreaction. Technical indicators such as the NHH and MRI improved short-term forecasting, uncovering exploitable inefficiencies in crisis-driven markets. The crisis exacerbated asymmetries in extreme price movements, driven largely by herding behaviour and irrational market reactions. These insights suggest that traders can capitalize on high-frequency tail dynamics for tactical advantages, while policymakers should focus on monitoring technical signals and mitigating systemic spillovers to enhance market stability.